open TELEMAC-MASCARET The mathematically superior suite of solvers

Bonjour,

Je travaille avec Matisse, et j'ai importé à peu prés 240 000 points de topo en formats *.xyz

Malgré la puissance de ma machine, lorsque je lance un calcul de maillage (avec des mailles de tailles importantes (500, 400, 200 mètres), Matisse plante et se ferme automatiquement.

Ma question est la suivante : Est ce qu'il existe une limitation en terme de nombre d'éléments pour Matisse ? Est ce que Matisse est capable de gérer un nombre de points aussi important ? Est ce que si j'essaye avec BlueKenue j'ai plus de chance de réussir ?

Je vous remercie d'avance pour toute information.
Cordialement,

Hi Yanis
Matisse is not perfect and sometimes crash. But as I know, 240 000 bathymetric points is supported.
One of the problem, is that with such number of data, Matisse takes a long time to compute a mesh ... but it's possible.
To increase the speed of computation, it's possible to only create a mesh with the boundary and the constraint lines (you can add few nodes for the refinement if you have one) but you set the elevation to 0 for all nodes. Once the mesh created, you could set the bathymetry directly in Fudaa or in Telemac (BOTTOM FILE).
With such method I generated a mesh of 800 000 nodes recently.
In your case, maybe your mesh size is too large, particularly if you have hard nodes on the boundary.
Does Matisse gives you an estimation of the number of nodes when you run the generation?
If NO, try with smaller mesh size.
See also the generated files in the BIN\Temp directory (defaultvertices) this could gives you an idea of the problem
About Blue Kenue, I never really test it. But it was used by the canadian (Hydro-Quebec and CHC) for very large area and with a high number of node in the final mesh (But it doesn't manage the bathymetric interpolation)

Good luck

Hi Yanis,

I don't have experience with Matisse so I'm not sure I understood whether your 200 000 points are considered during the meshing process? If so, I could understand why that computation is difficult.

I know we have worked with that many points in BlueKenue and have not had any problems. The work method we use is similar to that described by Christophe. We first generate a flat mesh (0 elevation) with constraint lines, boundaries, density zones and some hard points. We then interpolate our topo, bathy on that mesh with BK. I think the largest mesh we have generated so far is around 600 000 nodes. The largest .xyz file interpolated so far was several million points.

If your are interested to try, there is one other thread on assigning bathymetry with BlueKenue -> Telemac Forum : Assigning bathy. Sebastien put some links to the BK documentation. As far as I know, if you have a working flat mesh, you have a lot of software choice to interpolate bathymetry onto it afterwards (Fudaa, Telemac, BlueKenue, etc)


Hope this helps,

j

Hi Everyone,

Blue Kenue doesn't have any restrictions on the number of nodes or elements other than the amount of virtual memory available. With the release of the new 64 bit version there is practically no limit at all.

As for execution times, this depends on the number of constraints you supply. (ie. the number of break-lines, islands, density objects, etc...) Generally the mesher is reasonably fast though.
The attached image shows a fairly complex mesh of the Strait of Georgia (163 islands, element edge lengths from 40m to 4000m, 316,573 nodes and 551,759 elements). This mesh took 70 minutes to generate on a 2 year old PC.

Also as stated elsewhere, mapping bathymetry to your mesh is accomplished via the "Tools->Map Object" menu item.

Cheers... Martin